Very good at finding it is not the same as being right when it says yes
A positive result for Down syndrome in a low-risk pregnancy is right about 86 times in 100. For trisomy 18 it is right about half the time.
Two different numbers get muddled together here, and the difference decides how you should feel about a phone call.
The detection rate answers: if the baby has the condition, how often does the test say so? This number is genuinely excellent.
The positive predictive value answers a different question: if the test says yes, how often is it right? This is the number that matters when the phone rings, and it is the one nobody is given.
Detection first. Pooling 35 studies of singleton pregnancies:
| Condition | Detection rate | False-positive rate |
|---|---|---|
| Trisomy 21 | 99.7% (99.1 to 99.9) | 0.04% (0.02 to 0.07) |
| Trisomy 18 | 97.9% (94.9 to 99.1) | 0.04% (0.03 to 0.07) |
| Trisomy 13 | 99.0% (65.8 to 100) | 0.04% (0.02 to 0.07) |
Look at the range on trisomy 13. It runs from 65.8% to 100%, because the estimate rests on only 119 cases. The headline of 99% for trisomy 13 is not a solid number and should never be quoted without its range.
Now the number that matters. The largest prospective study with genetic confirmation of outcomes enrolled 20,194 women at 21 centres in six countries and had confirmed outcomes for 17,851 of them (88.4%).
| Condition | If your prior risk was low | If your prior risk was high |
|---|---|---|
| Trisomy 21 | 85.7% | 97.5% |
| Trisomy 18 | 50.0% | 81.3% |
| Trisomy 13 | 62.5% | 83.3% |
| All three combined | 74.3% |
Put in whole numbers: among low-risk women who get a positive result for Down syndrome, about 6 in every 7 results are correct and about 1 in 7 is wrong. For trisomy 18 in the same women, about 1 in every 2 positive results is wrong. Same test, same laboratory, same women. The difference is only how common the condition is.
This is also why nobody should quote you a positive predictive value from a laboratory's own confirmation caseload. Those figures describe a lab's workload, not a screening population.
ACOG's own summary table lists the screen-positive rate for this test as 2% to 4%, and notes that this
(includes inability to obtain results, which is associated with increased risk). So two to four women in a hundred get something other than a clean low-risk result.
"Is this a screening test or a diagnostic test?"
"What would the next step be, and how long would it take?"